As described in the previous issue, I tried to run the termite mound example script in a Jupyter notebook making only the minimum changes necessary for the script to run from a notebook. The default setting for the run_as_raster keyword is False, and the settings_termite_mound_example.yaml file contains "response_files: [['scratch/mounds_utm.shp']]". Running the example script with these settings results in training data that do not contain any termite mounds, and everything is classified as 'not mound'. Changing to run_as_raster=True and "response_files: [['scratch/mounds_utm.tif']]" avoids this problem.
As described in the previous issue, I tried to run the termite mound example script in a Jupyter notebook making only the minimum changes necessary for the script to run from a notebook. The default setting for the run_as_raster keyword is False, and the settings_termite_mound_example.yaml file contains "response_files: [['scratch/mounds_utm.shp']]". Running the example script with these settings results in training data that do not contain any termite mounds, and everything is classified as 'not mound'. Changing to run_as_raster=True and "response_files: [['scratch/mounds_utm.tif']]" avoids this problem.